Geographic Variation and Disparities in Total Joint Replacement Use for Medicare Beneficiaries: 2009 to 2017

Abstract

Background:

Little is known about how the geographic variation and disparities in use of elective primary total hip and knee replacements for Medicare beneficiaries have evolved in recent years. The study objectives are to determine these variations and disparities, whether Black Medicare beneficiaries have continued to undergo fewer total hip replacements and total knee replacements across regions, and whether disparities affected all Black beneficiaries or mainly affected socioeconomically disadvantaged Black beneficiaries.

Methods:

We used 2009 to 2017 Medicare enrollment and claims data to examine Hospital Referral Region (HRR)-level variation and disparities by race (non-Hispanic White and Black) and socioeconomic status (Medicare-only and dual eligibility for both Medicare and Medicaid). The outcomes were HRR-level age and sex-standardized total hip replacement and total knee replacement utilization rates for White Medicare-only beneficiaries, White dual-eligible beneficiaries, Black Medicare-only beneficiaries, and Black dual-eligible beneficiaries, and the differences in rates between these groups as a representation of disparities. The key exposure variables were race-socioeconomic group and year. We constructed multilevel mixed-effects linear regression models to estimate trends in total hip replacement and total knee replacement rates and to examine whether rates were lower in HRRs with high percentages of Black beneficiaries or dual-eligible beneficiaries.

Results:

The study included 924,844 total hip replacements and 2,075,968 total knee replacements. In 2017, the mean HRR-level total hip replacement rate was 4.64 surgical procedures per 1,000 beneficiaries, and the mean HRR-level total knee replacement rate was 9.66 surgical procedures per 1,000 beneficiaries, with a threefold variation across HRRs. In 2017, the total hip replacement rate was 32% higher for White Medicare-only beneficiaries and 48% higher for Black Medicare-only beneficiaries than in 2009 (p < 0.001). However, because the surgical rates for White and Black dual-eligible beneficiaries remained unchanged over the study period, the 2017 Medicare-only and dual-eligible disparity for White beneficiaries increased by 0.75 surgical procedures per 1,000 from 2009 (40.98% increase; p = 0.03), and the disparity for Black beneficiaries by 1.13 surgical procedures per 1,000 beneficiaries (297.37% increase; p < 0.001). The total knee replacement disparities remained unchanged. Notably, the rates for White dual-eligible beneficiaries were significantly lower than those for Black Medicare-only beneficiaries (p < 0.001 for both total hip replacements and total knee replacements), and fewer surgical procedures were conducted in HRRs with a higher density of Black or dual-eligible beneficiaries.

Conclusions:

Although the total hip replacement use for Medicare-only beneficiaries of both races increased, disparities for White and Black dual-eligible beneficiaries (compared with their Medicare-only counterparts) are increasing. Efforts to improve equity must identify and address both racial and socioeconomic barriers and focus on regions with high concentrations of disadvantaged beneficiaries.

Clinical Relevance:

Although total hip replacements and total knee replacements are highly successful surgical procedures for end-stage osteoarthritis, our findings show that, as recently as 2017, Black beneficiaries and those dual eligible for Medicaid (a proxy for socioeconomic status) are less likely to undergo these surgical procedures and that there is profound geographic variation in the use of these surgical procedures. This evidence is essential for the design and implementation of disparity-reduction strategies focused on patients, providers, and geographic areas that can potentially improve the equity in joint replacement care.

Total hip and knee replacements (“joint replacements,” hereafter) are highly successful surgical procedures given their ability to help to restore patients with advanced osteoarthritis to an optimal quality of life¹. However, not all patients who may clinically benefit from these surgical procedures undergo them. Although the prevalence of advanced osteoarthritis among Black individuals in the United States is similar to, if not higher than, that among White individuals^2,3, older Black individuals and individuals from lower socioeconomic groups undergo joint replacements less often than their counterparts^4,5. These disparities have been attributed to several factors, including patient preferences, which may be guided by inadequate information^5,6, provider biases while referring or recommending surgical procedures⁷, and geographic factors that determine availability and access to services^8,9.

Among geographic factors, both within-region mechanisms that result in lower joint replacement rates for Black individuals and between-region mechanisms such as a disproportionate number of Black individuals living in areas that have lower joint replacement rates for all have contributed to persisting variations^10,11. Studies that examined geographic differences in either hip replacements^12-14 or knee replacements^4,14 used data until 2001 and consistently found lower rates for Black individuals. Importantly, although these studies focused on racial differences, little is known about whether a Medicare beneficiary’s socioeconomic status mediated the geographic differences.

Our primary objective was to evaluate whether Black Medicare beneficiaries continued to undergo fewer elective primary total hip and knee replacements across regions in recent years, and whether these disparities equally affected all Black beneficiaries or mainly affected socioeconomically disadvantaged Black beneficiaries. We also examined whether these differences are attributable to the residential clustering of Black beneficiaries and socioeconomically disadvantaged beneficiaries in areas with lower utilization rates.

Materials and Methods

Data Sources and Study Cohort

We used the 2009 to 2017 Master Beneficiary Summary File (MBSF)-Base Segment¹⁵ to identify fee-for-service Medicare beneficiaries who were 66 to 99 years of age and were alive at the end of the calendar year for inclusion in the denominator (Appendix Exhibits S1 and S2). We used the geographic crosswalk files to map each beneficiary’s residential zip code to a Hospital Referral Region (HRR), which is a geographic health-care market for tertiary care^4,10,16. We used the 2009 to 2017 Medicare Provider and Analysis Review (MedPAR) files¹⁷ and the definitions developed and validated by the U.S. Centers for Medicare & Medicaid Services (CMS)¹⁸ to identify fee-for-service Medicare inpatient stays for elective primary total hip and knee replacements (Appendix Exhibits S3 and S4). We linked the MedPAR and MBSF files and aggregated them to the beneficiary level and then to the HRR level to generate annual counts of total hip and knee replacements for each HRR. We obtained HRR characteristics from the 2009 American Community Survey¹⁹, and the 2011 physician supply-side measures and the 2012 hospital supply-side measures from the Dartmouth Atlas of Health Care (https://www.dartmouthatlas.org/).

We limited our analysis to non-Hispanic White beneficiaries (“White beneficiaries” hereafter) and non-Hispanic Black beneficiaries (“Black beneficiaries” hereafter) because of their reliable identification from Medicare data²⁰ and because 97% of all total hip replacements and 96% of all total knee replacements were performed for these 2 groups (924,844 total hip replacements and 2,075,968 total knee replacements) (Appendix Exhibits S3 and S4).

Race and Socioeconomic Groups

We used the race variable in the MBSF files to identify the race of Medicare beneficiaries. These values are self-reported responses to questions about race and ethnicity that are obtained by the Social Security Administration. We defined socioeconomic groups based on a beneficiary’s enrollment in both Medicare and Medicaid programs. About 20% of Medicare beneficiaries who are poorer; have higher medical, functional, and behavioral needs; and have lower education qualify for Medicaid^21,22. We classified Medicare beneficiaries who were eligible for Medicaid for 12 months in a calendar year as dual-eligible beneficiaries to represent beneficiaries from the lower socioeconomic strata (Appendix Exhibit S5)²³.

Outcomes

The outcomes were HRR-level age and sex-standardized annual total hip replacement and total knee replacement rates expressed in thousands (referred to as “rates” hereafter). Rates were computed separately by race (White and Black), socioeconomic group (Medicare-only [higher socioeconomic status] and dual eligibility for both Medicare and Medicaid [lower socioeconomic status]), and their combinations (White Medicare-only, White dual-eligible, Black Medicare-only, and Black dual-eligible) using the direct standardization approach²⁴. This approach adjusted the total hip replacement and total knee replacement utilization for differences in age (5 age categories) and sex (male and female) composition of HRRs by using stratum-specific weights obtained from the national distribution of all Medicare beneficiaries (Appendix Exhibit S5).

Statistical Analysis

Descriptive Analysis

We examined the characteristics of the 306 HRRs, of Medicare beneficiaries residing in these HRRs, and of beneficiaries undergoing total hip or knee replacements in 2009 and 2017. We mapped the overall HRR-level rates for 2017 to determine the HRR-level variation in use.

Time Trends

To determine whether the mean rates for White Medicare-only, White dual-eligible, Black Medicare-only, and Black dual-eligible beneficiaries had changed significantly over time, we constructed separate multilevel mixed-effects linear regression models²⁵ for total hip replacements and total knee replacements with HRR-level random effects and an exchangeable covariance structure to account for clustering of observations within an HRR (Appendix Exhibit S5)²⁵. We modeled the age and sex-standardized rates of surgical procedures against the categorical race-dual eligibility variable (White Medicare-only, White dual-eligible, Black Medicare-only, and Black dual-eligible), a continuous specification of year, and interactions between these variables. The interaction term represents the change in HRR-level total hip replacement and total knee replacement utilization for each race-dual group over the study period. We also estimated the rates by race and socioeconomic status separately.

Between-HRR Differences

To determine whether HRRs with higher percentages of Black and dual-eligible beneficiaries were likely to have lower utilization of surgical procedures, we estimated separate multilevel linear regression models for total hip and knee replacements with HRR-level random effects and an exchangeable covariance structure. We modeled age and sex-standardized rates for each HRR against the baseline percentage of Black beneficiaries in the HRR (the 2009 percentage specified as quartiles), the baseline percentage of dual-eligible beneficiaries in the HRR (specified as quartiles), the interactions of each of these variables with a continuous specification of year, and HRR-level covariates (Appendix Exhibit S6). The interaction terms represent differences in HRR-level total hip replacement and total knee replacement utilization for HRRs with varying percentages of Black or dual-eligible beneficiaries and how these rates varied over the study period.

Sensitivity Analysis

We recalculated 2017 rates using Medicare beneficiaries with a diagnosis of osteoarthritis or rheumatoid arthritis as the denominator (identified using the MBSF-Chronic Condition Segment for 2017). We also reestimated the regression models using the Part D cost-share definition of dual eligibility, which represents the receipt of Part D premium and coinsurance subsidies for those who do not qualify for Medicaid (Appendix Exhibit S5)²³.

Results

Descriptive Statistics

HRR and Patient Characteristics

Of the 306 HRRs in 2017, the HRR with the highest percentage of White Medicare-only beneficiaries was Sun City, Arizona (92.44%) and that of White dual-eligible beneficiaries was Bangor, Maine (23.55%). New Orleans, Louisiana, had the highest percentage of Black Medicare-only beneficiaries (28.05%) and Black dual-eligible beneficiaries (11.59%) (Appendix Exhibits S6 and S7).

In 2017, 120,252 beneficiaries underwent 124,345 total hip replacements and 235,100 beneficiaries underwent 251,102 total knee replacements among a total of 26,713,917 White and Black Medicare beneficiaries (Table I).

TABLE I.

Characteristics of White and Black Medicare Beneficiaries Undergoing Total Hip or Total Knee Replacements, and All Medicare Beneficiaries, in 2009 and 2017^*

Characteristics	Total Hip Replacements†		Total Knee Replacements†		All Medicare Beneficiaries‡
	2009	2017	2009	2017	2009	2017
No. of patients or beneficiaries	83,949	120,252	204,012	235,100	24,557,784	26,713,917
No. of surgical procedures	86,308	124,345	220,267	251,102
No. of HRRs	306	306	306	306	306	306
Age§ (yr)	75.21 ± 6.32	74.17 ± 6.18	74.43 ± 5.97	73.66 ± 5.70	75.81 ± 7.53	75.04 ± 7.45
Female (%)	61.71	61.06	64.38	62.82	56.85	54.86
Black (%)	4.40	4.73	5.34	5.45	8.11	8.41
Dual eligibility (%)	5.01	3.96	6.18	4.51	10.20	8.50
Beneficiary category (%)
White Medicare-only	91.43	92.03	89.67	90.88	83.81	84.87
White dual eligibility	4.17	3.24	4.99	3.68	8.08	6.72
Black Medicare-only	3.56	4.01	4.15	4.61	5.99	6.62
Black dual eligibility	0.84	0.72	1.19	0.83	2.12	1.79
Region (%)
Northeast	19.10	19.30	15.64	16.82	19.24	18.68
Midwest	28.13	25.02	28.28	25.30	24.65	22.62
South	34.82	35.83	40.39	40.48	39.97	40.21
West	17.95	19.85	15.69	17.40	16.14	18.49

^*The cohort includes fee-for-service Medicare beneficiaries between 66 and 99 years of age who were alive at the end of 2009 or 2017.

^†The details of this cohort definition are presented in Appendix Exhibits S3 and S4.

^‡The details of this cohort definition are presented in Appendix Exhibits S1 and S2.

^§The values are given as the mean and the standard deviation.

Geographic Variation in HRR-Level Rates

In 2017, total hip replacement rates ranged from 2.01 surgical procedures per 1,000 beneficiaries in Alexandria, Louisiana, to 7.55 surgical procedures per 1,000 beneficiaries in Traverse City, Michigan (mean [and standard deviation], 4.64 ± 1.09 surgical procedures per 1,000 beneficiaries), and total knee replacement rates ranged from 4.90 surgical procedures per 1,000 beneficiaries in Lebanon, New Hampshire, to 16 surgical procedures per 1,000 beneficiaries in Ogden, Utah (mean, 9.66 ± 1.87 surgical procedures per 1,000 beneficiaries), indicating a threefold variation (Appendix Exhibit S8).

In 2017, the mean total joint replacement rates were 4.77 total hip replacements and 9.93 total knee replacements, both per 1,000 beneficiaries, for White beneficiaries, and 2.80 total hip arthroplasties and 6.13 total knee arthroplasties, both per 1,000 beneficiaries, for Black beneficiaries (in HRRs with at least 1% Black beneficiaries) (Fig. 1). The total joint replacement rates for Black beneficiaries were above the national mean of HRR-level rates for White beneficiaries in only 19 HRRs (7.92%) for total hip replacements and 11 HRRs (4.58%) for total knee replacements.

Fig. 1.

Age and sex-standardized hip and knee replacement rates for White and Black beneficiaries in 2017, based on analysis of 2017 MedPAR and MBSF and Dartmouth Atlas of Health Care Zip Code-HRR crosswalk files. White and Black beneficiaries were identified using the race variable in the MBSF files. The utilization rates were computed using the direct standardization method. In HRRs in which Black beneficiaries constituted <1% of the Medicare population (n = 66), the HRRs were shaded gray. The maps were prepared using QGIS 3.4 (QGIS Development Team) and the categorization of rates was done using the natural breaks methodology in QGIS. The categorization of rates for Black beneficiaries was set to match those for White beneficiaries.

Time Trends

The overall mean HRR-level total hip replacement rate in 2017 was 32.66% higher (p < 0.001) in 2017 at 4.59 surgical procedures per 1,000 beneficiaries compared with 3.46 surgical procedures per 1,000 beneficiaries in 2009. Also, the overall mean HRR-level total knee replacement rate was 2.40% higher (p < 0.001) in 2017 at 9.40 surgical procedures per 1,000 beneficiaries compared with 9.18 surgical procedures per 1,000 beneficiaries in 2009.

In 2009, the total hip replacement rate for White Medicare-only beneficiaries (3.71 surgical procedures per 1,000 beneficiaries) was almost twice the rate of other groups (1.88 for White dual-eligible beneficiaries, 2.21 for Black Medicare-only beneficiaries, and 1.83 for Black dual-eligible beneficiaries; p < 0.001), with similar patterns for total knee replacement (Fig. 2) (the regression estimates appear in Appendix Exhibit S9). In comparison with 2009, the 2017 total hip replacement gap between White Medicare-only and Black Medicare-only beneficiaries remained unchanged (2009 gap of 1.49 and 2017 gap of 1.63, p = 0.68), that for White Medicare-only and White dual-eligible beneficiaries increased by 0.75 surgical procedures per 1,000 beneficiaries (2009 gap of 1.83 and 2017 gap of 2.58, with a 40.98% increase in the gap; p = 0.03), and that for Black Medicare-only and Black dual-eligible beneficiaries increased by 1.13 surgical procedures per 1,000 beneficiaries (2009 gap of 0.38 and 2017 gap of 1.51, with a 297.37% increase in the gap; p < 0.001). Thus, total hip replacement disparities increased because, although rates for White and Black Medicare-only beneficiaries were significantly higher in 2017 than in 2009 (32% higher for White Medicare-only beneficiaries [p < 0.001] and 48% higher for Black Medicare-only beneficiaries [p < 0.001]), the rates for White and Black dual-eligible beneficiaries did not change over time. There were no changes in total knee replacement use for each group or disparities between groups. Notably, White dual-eligible beneficiary rates were significantly lower than Black Medicare-only beneficiary rates for both total hip replacements and total knee replacements (p < 0.001) through most of the study period. In sensitivity analyses, the White compared with Black gap in 2017 remained unchanged for total hip and knee replacements, and the Medicare-only compared with dual-eligible gap increased significantly compared with 2009 (Appendix Exhibit S10).

Fig. 2.

The mean HRR-level age and sex-standardized hip and knee replacement rates by combinations of race and income groups based on analysis of 2009 to 2017 MedPAR and MBSF. The lines represent adjusted estimates derived from multilevel mixed-effects linear regression models with HRR-level random effects and an exchangeable covariance structure (Appendix Exhibit S9). Post-estimation averages were used to plot the time trends. The superscript “a” denotes the differences in rates (surgical procedures per 1,000 beneficiaries) between Medicare-only and dual-eligible beneficiaries for each racial group in 2009 and 2017.

Between-HRR Differences

In 2017, the mean total hip replacement rates in HRRs with high percentages of Black beneficiaries were 6.37% lower for quartile 4 (p = 0.01) than quartile 1, and, compared with quartile 1, the mean total hip replacement rates in HRRs with high percentages of dual-eligible beneficiaries were 3.67% lower for quartile 2 (p = 0.03), 7.35% lower for quartile 3 (p < 0.001), and 14.08% lower for quartile 4 (p < 0.001) (Table II, Appendix Exhibit S11). The differences in total knee replacement rates between HRRs exhibited similar patterns.

TABLE II.

Mean HRR-Level Age and Sex-Standardized Hip and Knee Replacement Rates in 2017 by Quartiles of the Percentage of Black or Dual-Eligible Medicare Beneficiaries^*

	By Quartiles of Black Beneficiaries				By Quartiles of Dual-Eligible Beneficiaries
	Hip Replacement		Knee Replacement		Hip Replacement		Knee Replacement
	Adjusted Rate†	Difference	Adjusted Rate†	Difference	Adjusted Rate†	Difference	Adjusted Rate†	Difference
Quartile 1	4.71	—	9.59	—	4.90	—	10.02	—
Quartile 2	4.69	−0.42%	9.66	0.73%	4.72	−3.67%‡	9.78	−2.40%
Quartile 3	4.56	−3.18%	9.33	−2.71%	4.54	−7.35%§	9.31	−7.09%§
Quartile 4	4.41	−6.37%‡	9.00	−6.15%‡	4.21	−14.08%§	8.46	−15.57%§

^*The table presents the 2017 adjusted estimates of hip and knee utilization rates by quartiles of the percentage of black or dual-eligible Medicare beneficiaries in the HRR. The estimates were derived from multilevel mixed-effects linear regression models with HRR-level random effects and an exchangeable covariance structure that used data from 2009 to 2017 (Appendix Exhibit S11).

^†The values are given as the rate per 1,000 surgical procedures.

^‡Significant at p < 0.05.

^§Significant at p < 0.001.

Sensitivity Analysis

The findings from sensitivity analyses that focused on beneficiaries with a diagnosis of osteoarthritis or rheumatoid arthritis (results not presented) and from the Part D cost-share data-based definition of dual eligibility were consistent with the main findings (Appendix Exhibit S12).

Discussion

We examined the geographic variation in total hip replacement and total knee replacement rates across race-socioeconomic groups of Medicare beneficiaries from 2009 to 2017. We found that the overall use of joint replacements varied threefold across HRRs, and the rates for Black beneficiaries were above the national mean of HRR-level rates for White beneficiaries in only 5% to 8% of HRRs. The total hip replacement rates for White and Black Medicare-only beneficiaries were significantly higher in 2017 than in 2009. However, the 2017 rates for their dual-eligible counterparts were not significantly higher than 2009 rates, thereby substantially widening disparities between Medicare-only and dual-eligible beneficiaries for each race in 2017 compared with 2009. The total knee replacement disparities remained unchanged. Notably, in 2017, White dual-eligible beneficiaries continued to have lower total hip replacement and total knee replacement use than both White and Black Medicare-only beneficiaries, and Black dual-eligible beneficiaries had the lowest use of these surgical procedures, highlighting the critical role of a beneficiary’s socioeconomic status in influencing surgical utilization, and the disadvantage that dual-eligible beneficiaries face regardless of their racial identity. Furthermore, surgical rates continued to be lower in areas with a high concentration of Black or dual-eligible beneficiaries. Together, these findings suggest that efforts to reduce racial disparities have, at best, achieved partial success in improving total hip replacement use for Black beneficiaries from the higher socioeconomic strata and no success for Black or White beneficiaries from lower socioeconomic strata or for total knee replacements. By elucidating disparities in both total hip replacements and total knee replacements, we generate important evidence for action. A continued commitment to measurement, evaluation, and refinement of existing approaches that, in particular, address the social determinants of health is valuable for narrowing these gaps^26-28.

Our findings confirm inferences from previous geographic-level studies^4,12-14 that used data up to 2001 and patient-level studies^29,30 that used data up to 2008 and demonstrated disparities in the use of joint replacements. Our study fills an important gap in the literature by providing an update about the use of joint replacements for White and Black Medicare beneficiaries in recent years and makes an important contribution by examining how a beneficiary’s socioeconomic status (as measured by dual eligibility) mediates the use of joint replacements. These disparities have potential implications for the quality of life of disadvantaged Medicare beneficiaries³¹.

The larger overall increase in total hip replacement rates and the increases in total hip replacement rates for White and Black Medicare-only beneficiaries are consistent with several explanations about the growth in these surgical procedures, including the aging of the population, increase in the prevalence of obesity, changing indications for a surgical procedure, and increase in physical activity patterns³². In this current study, we did not note exponential increases in total knee replacements that studies have previously noted³², mainly because of the differences in the study designs, and because some of the increase in previous studies may have occurred among the individuals who were <65 years of age³³. The relatively stable disparities for total knee replacements may be attributed to the overall stability of total knee replacement rates during the study period, and larger disparities at the start of the study period compared with total hip replacements. Anecdotally, because total hip replacements may be considered surgically more complex than total knee replacements by general orthopaedic surgeons, and with an increasing prevalence of obesity among racial minorities and those from lower socioeconomic strata³⁴, surgeons may be reluctant to operate on these patients, thereby increasing total hip replacement disparities to a greater extent.

Although overuse by White beneficiaries⁴ may be one of several explanations for the Black-White differences that we note in our study, the lower use by Black beneficiaries may also be explained by their lower preferences for these surgical procedures. However, because these preferences are shaped by “truncated opportunities or historical circumstances” that Black beneficiaries encounter⁵, the differences that we observe are markers of systematic disparities. Furthermore, the reliance that Black beneficiaries have on alternative forms of treatment³⁵, their uncertain expectations with regard to the surgical course³⁶, their cultural or language barriers while navigating the system¹³, their fewer family or friends who may have undergone these surgical procedures³⁷, or their families and friends who have had bad outcomes are some of the many reasons why Black beneficiaries may be less willing to undergo these surgical procedures even when recommended by their physicians or during educational interventions^38,39.

The lower use of joint replacements among lower-income and dual-eligible beneficiaries has been previously demonstrated^4,40. Despite Medicare being a federal insurance program that covers most older adults, the ability to pay for supplemental gap insurance or out-of-pocket expenses⁶, a lower perception of benefits among lower socioeconomic groups⁴¹, the limited access to primary care physicians, and the role of health behaviors and regional factors⁴² may partly explain the lower utilization of joint replacements by dual-eligible beneficiaries.

HRRs with high proportions of Black and dual-eligible beneficiaries performed fewer joint replacements for all Medicare beneficiaries in comparison with other HRRs. These between-HRR differences were previously noted by Skinner et al. using measures of racial segregation to demonstrate lower use of joint replacements⁴. Thus, policies that focus on optimizing the use of surgical procedures in HRRs with lower rates are likely to reduce the between-HRR variation and improve care for all beneficiaries regardless of their race or socioeconomic status.

In recent years, payment reform has focused on the quality and cost of joint replacements. However, these changes may incentivize hospitals to selectively avoid more medically complex vulnerable patients in order to score higher on quality metrics linked to reimbursements⁴³. Payment reform may also disadvantage hospitals who treat disproportionate volumes of these patients⁴⁴. In the absence of specific metrics that reward hospitals for reducing gaps in the use of these surgical procedures or risk-adjustment methods that level the playing field for safety-net hospitals, it is unlikely that the racial and socioeconomic disparities can be eliminated.

Our study had several limitations. First, because our research was focused on the geographic level, it did not adjust for patient-level confounders, including differences in health. However, sensitivity analyses that were limited to beneficiaries with osteoarthritis and that controlled for HRR-level covariates such as income and education produced findings that were consistent with the main findings. Second, elective joint replacements are preference and culture-sensitive surgical procedures and our analysis was based on national claims and enrollment data, which do not capture these constructs. Next, our findings are sensitive to the methodology used to define the cohort (Appendix Exhibits S1 through S4). Because our study focused on fee-for-service Medicare beneficiaries who were eligible for or underwent elective primary total hip replacements or total knee replacements, we applied several exclusion criteria such as exclusion of Medicare Advantage or Part C beneficiaries, those <66 years or >99 years old, those who died during the year (from MBSF files), and those who underwent joint replacements for fractures, neoplasms, or mechanical complications or underwent removal, revision, or resurfacing or partial joint replacements (from MedPAR files). Thus, our findings are mainly generalizable to Medicare beneficiaries covered by fee-for-service (Part A) Medicare who undergo elective primary total hip or knee replacements. Finally, we chose dual eligibility as a proxy for socioeconomic status because of previous literature that has shown that these individuals are poorer and have numerous impediments^21,22,45 to full social participation. Moreover, because dual eligibility reasonably captures both poverty and fuller insurance coverage, assessing racial disparities within dual-eligibility status is a reasonable test of the hypothesis that racial disparities result from more than poverty and unaffordable out-of-pocket costs not covered through Medicare.

In conclusion, our study demonstrates that significant geographic variation and racial and socioeconomic disparities in the use of joint replacements persisted during the 2009 to 2017 period. Efforts directed toward beneficiaries of lower socioeconomic strata and those residing in regions with high concentrations of Black or dual-eligible beneficiaries are imperative to achieve equity in the use of joint replacements.